Clinical and applied research - grounded theory - scholarly communication



--

Copyright © 2001 by Ken Friedman.

All rights reserved.

This text may be quoted and printed freely with proper acknowledgment.

--



Thanks to Terry, Rob, Chris, and Wim for their thoughts and comments.

I want to address three sets of issues here.

1) The distinctions between clinical and applied research,

2) Aspects of grounded theory and inference.

3) The relation between these and scholarly communication.

I am comfortable with the distinctions that Terry raises.

I am less comfortable with Rob's response. It conflates the very
aspects of applied and clinical research that we are trying to
clarify.

Rob writes, "I think that applied research in design is research
where the findings of the research are applied to the process of
designing something. This type of research is what I get involved in.
For example researching a group's color preferences and then applying
the color to the object which is sold to that group."

This definition of applied research would have been used more widely
in the past that it would be today. It is possible that this would
still be used this way in industry precisely because industrial
research specialists at Rob's level tend to have graduated two or
three decades back. At that time, people frequently labeled BOTH
applied and clinical research under the rubric of applied research.

The difficulty is that when we place both forms of research under the
label of applied research. When clinical and applied research are
both aggregated under single the label of applied research, we cannot
distinguish between a general range of applications or a classes of
cases and the specific application instantiated in a single case.

The distinction offers us several important values. The first value
of the distinction lies in helping us to distinguish among the
different kinds of research methods top is used. Another is the help
it gives us in distinguishing among different pedagogical issues
implied in each form of research. The third is that fruitful research
is anchored in clarity of distinction.

Distinguishing among basic, applied, and clinical research enables us
better to understand methodological issues. [See note 1.]

Developing these kinds of distinctions is an important aspect of a
progressive research program for design research.

It is important to be clear on terminology. Even though the findings
of clinical research are applied, this does not mean that clinical
research is applied research.

Inquiring into the specific color preference of a particular group
constitutes clinical research because it involves research on a
specific instance or case.

The findings of clinical research findings are nearly always applied
in specific instances or cases.

In contrast, applied research generates findings that can be applied
in many cases.

The methods that any designer uses to determine color preferences are
a method that has probably been established by applied research in
design research methods.

The specific research done to establish the specific preferences of a
client group is clinical research.

This can be compared to medical research.

Basic research determines the cause of a disease or condition.
Finding out how the disease works and spreads is often basic research.

Finding out how to diagnose the disease, and establishing the indices
and medical factors around the disease involves applied research.
Examining the spread of a disease in a specific environment probably
lies at the border of applied research and clinical research. It can
also contribute knowledge to basic research.

Diagnosing a disease or condition in a specific patient is clinical
research. Finding an applicable cure for that patient involves
clinical research and it involves the application of clinical and
applied findings.

While clinical research generates findings that are applicable to one
case, they may also yield information that contributes to a larger
range of cases.

This leads to some of the issues that Chris raises.

Many aspects of product design involve clinical research.

Terry offered a specific case in engineering design as an example of
clinical research:

"Clinical research: Resolving a piston ring wear problem in a Volvo
XXXX - resulting in a solution to a piston ring wear problem specific
to that model of vehicle and particular circumstances."

Chris responds to Terry's Volvo example:

"The research activity is comparable to the kind of experimental or
analytical work which arises in almost every design discipline when
the designer is seeking to innovate rather than apply a proven
formula. We have had more than one debate about this and colleagues
have been forthright about the importance of differentiating between
the creative output of an artist or designer and the contribution to
knowledge made by the researcher (even if they sometimes co-exist in
a single piece of work)."

Clinical research may well involve innovation, but it generally does
not contradict established findings.

The Volvo piston ring development project is research. Research takes
place in industry and in the design studio as well as in the context
of academic research.

Some of the best and most advanced research being done in many fields
takes place in industry and in the studio. Along with leading-edge
research, there is also a lot of tinkering, and plenty of mistakes.

The combination of wants, needs, environment, and a ruthless weeding
out process makes industrial application a powerful laboratory.

This is both good and bad. It is good because it produces so many
results. It is bad because the process is often wasteful. Part of the
laboratory process is imposed on designers and companies by a process
that resembles natural selection in evolutionary theory rather than
by a process selected by them through effective research.

Some form of research goes into every new product. The research may
be good or bad, but every company engages in some form of research
before committing funds and resources to a new product.

Every new product is also designed. The design may be good or bad,
but there is some form of design.

The issue that concerns those of us engaged in design and design
research is finding better ways to engage in the design process and
the research process.

It is important to recognize that the bad methods of research and
design we have all seen remain methods of research and design,
however bad.

When a design team puts a dozen prototypes on a table so that six
consumers can choose their favorites, it is a form of research.

When the design team puts the tanked favorites on another table to
ask an executive committee to decide which is most likely to succeed,
that is a form of research and development.

These are inadequate, but they constitute forms of research and development.

When the marketing chief who heads a new product team leaves a color
choice up to his art student nephew, he makes a design choice.

One way that hungry design firms work is to generate hundreds of
ideas, submitting the sketches to clients, and selecting the
proposals clients like best for further work. It is not good, it is
not the way serious professionals work, but many of us have seen it
happen Š or seen its results.

These forms of choice lead to an evolutionary process, where the
market serves as an environmental niche within which a form of
natural selection chooses the product best suited to the niche.

Evolution functions as a form of development process. Much like the
process of evolution in natural, random development and selection
work effectively at a high price in failed developments and extinct
lines.

The evidence of new product failure is clear. Mansfield et al. (1971:
57) studied new product ideas in different stages of development from
proposal to rollout to results. Of new product ideas that moved
beyond the proposal stage, they found that 57% achieved technical
objectives, 31% entered full-scale marketing, and only 12% earned a
profit.

According to some experts, over 80% of all new products fail when
they are launched, and another 10% fail within five years (Edwards
1999, Lukas 1998, McMath 1998). Planning and study can never yield
perfect results. Even so, a robust research process linked to a
robust design process can improve the odds.

Excellent designers enhance the success rate of products by using
robust research methods and an articulated problem-solving process.

Part of what we are doing on this list is exploring the range of
robust research methods, and sharing ideas on such aspects of design
research as methodology and distinctions.

Developing better methods and distinctions for research is one
purpose for academic research. Science and scholarship operate under
a different set of constraints than practicing designers in industry
and business. This gives university-based researchers the opportunity
to explore issues and make discoveries. It gives them the chance to
learn why things work, rather than simply finding out that they work.
In finding out why things work, we also learn better how to apply our
working knowledge. In essence, discovering why something works also
leads to a range of better techniques and methods in how it works.

Where working designers under pressure for immediate results are
often limited to clinical research, scholars and scientists engaged
in design research are able to pursue basic and applied questions as
well.

Industrial constraints, time constraints, and budget constraints
often require companies and design groups to make inadequately
informed choices. In Herbert Simon's term, they satisfice.

One of the reasons effective executives hire skilled designers with a
solid record of accomplishment is that they have learned to make
reasonably good decisions in a reasonable number of cases. They have
a reasonable batting average, and it sometimes seems to make sense to
bet on the right design firm to produce a hit.

Nevertheless, this is increasingly unacceptable. The idea of design
research is that we should eventually be able to find ways to do
better and waste fewer resources. Anthropologist Bryan Byrne comments
that as companies begin to commit millions of dollars to product
launches, they can no longer afford to make haphazard choices on
design parameters. This requires sophisticated research, and this, in
turn requires methodological clarity.

Thus, it becomes useful to distinguish between basic, applied, and
clinical research. These are not merely academic understandings, but
a guide to choices and a framework for understanding.

As we move toward that goal, I sometimes feel we are today where
physics was in 1895 or 1900 relative to where physics is now. We are
not yet where physics was in 1905. Great days are ahead, but we have
a lot of sorting out and clarification to do.

Chris also raised the issue of grounded theory and induction. I will
touch on this without answering fully.

Grounded theory (Glaser 1978, 1992; Glaser and Strauss 1967; Strauss
and Corbin 1990) builds emergent theory from a foundation in
individual cases. The explanation of how this is done and why it
works is a far lengthier discourse than I am ready to attempt today.

Grounded theory was an important contribution to social research. It
builds on the same foundation as the ideas of Herbert Blumer and
symbolic interactionism, and it is anchored more deeply in the
traditions of Chicago sociology going back to George Herbert Meade.

Grounded theory does not elaborate theory from a single case. Single
cases provide data and ideas.

Grounded theory provided a method for accumulating, tabulating, and
organizing ideas in a structured system across a range of cases.
Theory emerged from the range of data.

Grounded theory involves three issues that are not always clearly understood.

The first is that nearly anything that grounded theory can do can be
done with symbolic interactionist theory. What makes grounded theory
useful is that it offers a far better structured and more systematic
way of doing it.

Both methods require theoretical sensitivity, but grounded theory
offers a better opportunity to grasp and manage data.

Some years ago, one of my thesis students was working on a problem
that seemed to me to call for symbolic interactionist methods. I set
him to work with Blumer, and he dug in. After about three months, he
came to me complaining that he was swimming in a sea of data that he
found difficult to bring into reasonable order.

I gave him Glaser and Strauss, and suggested he use grounded theory
to tame his data.

He went to work. About a month later, he came back, complaining that
grounded theory was so systematic and demanding that the task had
become a terrible process of the coding and structuring of data that
grounded theory requires.

My answer was that his task required one or the other. People, with a
great tolerance of ambiguity, a willingness to swim in data until a
theoretical structure emerges, and a lively sense of social and
cultural evolution across historical data sets do better with
symbolic interactionism. People with a greater tolerance for routine,
structured activity, and pattern recognition through coding systems
and organized schemata do better with grounded theory.

Neither is better. They are different.

Second, grounded theory has a specific virtue. It renders qualitative
social data as coded quanta. The great virtue in this system of
codification is that it answered a political problem in social
science research.

Herbert Blumer, the great sociologist and a past president of the
American Sociological Association, often complained that American
sociology suffered from two great flaws. One was a tendency to
quantify, and a belief that that which could not be quantified was
meaningless for sociological research. The other was a tendency to
substitute sophisticated methods for robust conceptualization (Blumer
1969: 1-60, 140-170).

While a reasonable position of naturalistic inquiry and a sensible
realist approach to the philosophy of science suggest tat not every
aspect pf empirical reality is best handled through quantitative
research methods. In the 1960s and 1970s, however, many American
sociology departments seemed to operate on the belief that only
quantifiable social data were worth studying.

The grand political stroke of Glaser and Strauss was this. By shaping
a coding technique for qualitative data, they rendered the
qualitative quantifiable. In this way, they freed a generation of
sociology researchers from the statistical manipulation of true but
trivial data.

Grounded theory and the coding system its employs have other
purposes, quite valid. For example, grounded theory methods can be
used for many of the same purposes that content analysis serves.

It is precisely because grounded theory allows coding and comparison
across a range of cases aggregated into a class that it permits
theoretical structures to emerge. Some theoretical structures may
emerge in a rich data set accumulated in one, large case. An example
of this would be a series of many interviews in a large organization.
Often, grounded theory structures data from many comparable cases.

This is such a quick, streamlined answer that I am not going to
suggest this accounts for all the possibilities. It does not.
Nevertheless, one case alone with too small a data set will not yield
emergent theory. Emergent theory arises from a data set that is rich
enough to form the ground against which theory can emerge. Putting
this into the language of gestalts, theory is the figure that emerges
against the ground. This is one reason that grounded theory is called
grounded theory.

Third, a most important, induction along is insufficient for
theoretical discovery. There is and must be an interplay between
induction and deduction, between the particularly and the general.

This is such a deep issue in the philosophy of science that I will
say no more at this point.

Induction functions in the logic of discovery.

An issue in design research that lies outside the frame of this
thread involves confusion between the logic of discovery and the
logic of justification.

One of the annoying phenomena of the past few years is the tendency
of design research students - and a few professors -- to misuse C. S.
Peirce's concept of abduction. Abduction, properly speaking, is a
generative technique. Abduction functions within the logic of
discovery, but it does not function in the logic of justification.
Abduction points to possibilities that must be tested and
demonstrated through justification, proof, or absence of disproof.

The failure to understand abduction leads to many silly comments by
design scholars with inadequate research training (e.g., Wood 2000:
52)

The problem is that too many design scholars who discover a concept
digest it half-understood or misunderstood without taking the time to
read deeply. Instead, they wave a reference at the concept as thought
something printed in a book proves their point. It is a form of
superstitious research using references as talismans. The reason I
address it here is to contrast abduction from single cases with the
larger ground required by grounded theory.

Peirce himself was a scientist and mathematician as well as a
philosopher. He explained his view of scientific method in an
important article (Peirce 1877: 1-15) stating that empirical reality
was the determining factor in justified belief. Much of his work
involved an analysis of the forms of inference and their
justifications.

He distinguished between three main forms of inference, abduction,
deduction, and induction, and he argued for a method that in today's
terms would be labeled the hypothetical-deductive method (Hilpinen
1995: 566-567).

There has been serious debate about the status of abduction. Some see
it as common sense (e.g., Mautner 1996: 1). Others (f.ex., Bunge
1999: 5) view abduction as a mere intellectual fashion, a short-lived
curio hardly worth discussing. I tend toward Mautner's view, but the
way that Peirce is so often cited and misused in design research
would certainly justify Bunge's disdain for the trendy misuse of
intellectual fashions.

The reason I discuss abduction here is to clarify the basis of grounded theory.

In building an appropriate ground for emergent theory, grounded
theory develops a foundation for inductive discovery. Among the
inductive patterns, abduction will reveal potential candidates for
theory construction. These are then examined and tested in a
deductive system.

I have to disagree with Chris where he writes, "Terry's other example
from social science is also quite specific, but it is about knowledge
rather than workable solutions to problems, although they may follow.
The principle of grounded theory allows for generalizable theory to
grow from situation-specific theory and even suggests that this may
be a more realistic route to reliable knowledge than the top-down
approach implied by basic research."

Generalizable theory can grow from specific cases, but the process is
an emergent process based on multiple cases. Developing theory
requires the interplay of induction and deduction together with
testing or criticism of some kind.

It is a mistake to see basic research as "top down." The empirical
world is the foundation of all research in field such as social
science, design science, or even chemistry or physics. Only
axiomatized research fields such as geometry or mathematics can
proceed from pure deduction.

Basic research in social science, natural science, or design science
is a more subtle matter.

Chris's example of distortion during the firing of large ceramic
castings is a case in point. He notes that, "Generations of craftsmen
have learned to predict the effects of distortion and minimize the
number of reworks required to get the mould right."

In essence, this is another way of saying that these generations of
artisans have accumulated a wealth of experience through clinical
research. Their research knowledge is tacit knowledge. That is, it is
embodied knowledge. No one has yet been able to translate this
embodied knowledge articulate information. Thus, it remains clinical
research but it has not yet become applied research.

These clinicians have nevertheless accumulated a rich inductive
experience. While we do not yet know how to translate their
knowledge, they know how to use their knowledge. This is not
knowledge of one case, or of several cases, but of generations of
cases. When you consider that a ceramics master must fire several
times a day, one master craftsman must accumulate the experience of
eight to ten thousand fittings over the course of a full working
life, perhaps more. The craftsmen who have fired these large ceramic
castings go back at least to the first large-scale firings in the
late 1700s and early 1800s.

If you do a little back-.of-the-envelope math, you will immediately
see that this is precisely the kind of massive foundation of
experience that should give rise to emergent theory.

In the short run, it may not be worth the efforts of one doctoral
candidate to render this knowledge explicit. That depends on the
specific doctoral student. This does not mean it cannot be done.
Poincare almost discovered the theory of relativity. Einstein did.

The essence of knowledge creation in the industrial setting involves
navigating the cycle between explicit and tacit knowledge. We apply
intellectual effort to tacit knowledge until we are able to render it
explicit. By rendering tacit knowledge explicit, we move it from the
form in which we alone know it into a form that allows others to know
it.

Then, it is necessary to render explicit knowledge tacit again. By
this, I mean that we work with explicit knowledge until it is once
again embodied and made real in our daily practice.

Chris's craftsmen have the knowledge. The challenge of the researcher
in this case involves rendering the knowledge explicit and making it
usable by others - or making it accessible in a CAD-CAM system.

CAS-CAM systems are less than thirty years old. My prediction is that
this knowledge can be rendered explicit given time. It must have
taken over two centuries for these generations of craftsmen to create
and embody their knowledge. They posses the emergent theory, even
though they cannot yet put it into words.

It seems to me that a patient researcher - or several of them -
willing to struggle with this problem will develop articulate
emergent theories in far less time than the craftsmen have required
to create and embody the knowledge they possess.

Now, Chris's note actually has a cheerful ending with several useful lessons.

"More recently CADCAM engineers have been able to describe the
distortion effects in the form of pseudo materials properties that
might be understood by a commercial CAD modeling system and give
reliable enough predictions of distortion," he writes. Obviously,
human beings have already learned a lot based on a wide range of
cases.

He continues, "All of this knowledge came from individuals'
experiences of specific cases - getting a specific toilet bowl to
fire true by the end of next week - but has become very useful in
general terms."

He goes on, adding that, "somebody did start doing a PhD aimed at
developing true CAD modeling of the actual behavior of the clay but
realized it would be an enormous effort, which might not get a result
. . . "

This is a risk in all research. Many times, we start on a project
only to discover that it has already been done. Sometimes we discover
that there are no results. Other time, we discover that we were
wrong. Sometimes, we even discover that we were right and the
knowledge is a modest yield for the investment.

Chris himself points to the modest yield position in concluding, "and
which would be of very limited benefit given the specialized nature
of the problem so they went back to their company and worked on
improving the development process and existing CAD tools instead."

These are the risks of science.

We engage in research. Winning new knowledge is hard. When we win
knowledge, humanity gains the benefit of our investment. This is why
research is a social good. This is why universities are
not-for-profit agencies. They do not meet the profit criteria that
shape the decisions of industrial corporations. They are not expected
to do so. They produce a social good, and this yield is measured in
different ways.

With this, I will offer a last thought on the distinctions of research.

Working designers operate under the harsh constraints of a need to
turn profit on every project. They are constrained, therefore, to
clinical research in most cases, and applied research in only a few.
When they do engage in applied research, working designers are often
required to keep their findings secret. Thus, even their applied
research yields only specific clinical results, and those results are
not disclosed.

Design research has the discovery of many kinds of knowledge as its
goal. We seek to create pure knowledge for humankind. We seek applied
and technical knowledge for design, design studies, design science,
and for professional designers. We seek clinical knowledge for the
improvement of professional practice.

All these are good.

I raised these issues in the context of a thread on scholarly
communication. Once again, I will point back to that thread.

My point is three-fold.

The first point is that we need all these kinds of knowledge.

When designers are active in design research in the university
setting, they must consider basic and applied issues as well as the
clinical issues with which they are already familiar. If a designer
chooses to undertake design research, he or she must meet the
standards of research as well as the standards of professional
practice.

The rest of us are active in developing ideas, issues and
contributions that will improve the field of design and improve the
field of professional design practice.

It is expected that designers active in design research should also
be concerned with the full range of these issues, and not merely
studio practice and demonstration of personal results.

The second point is that this requires articulate scholarly communication.

Whatever results we get must be shared with the field to be meaningful.

When someone works to gain personal knowledge for him or herself,
that is study. Someone working on personal knowledge is a student.

Someone who shows that he or she can apply personal knowledge to
solve a problem demonstrates. A person who demonstrates personal
knowledge is a professional.

When someone explains how he or she gained knowledge, what the
knowledge is, how it works, and how to apply it, the act of
explanation transforms study and professional skill into research.
The person who does this is a scholar or a scientist.

The third point is that these areas are never clear-cut. None of
these issues are exactly black and white, but rather, all these
issues lie on a continuum of understanding and development.

It is for precisely this reason that it is useful to understand the
distinctions among basic, applied, and clinical research, and to
reflect on them in our research activities.

-- Ken Friedman




[Note 1]

Methodology is the comparative study of research methods. Methodology
is one area of method study in research. It is distinct from
methodics, the collection of research methods used in any given
field. It is distinct from a specific method. It is distinct from
technique, the application of method.



References


Blumer, Herbert. 1969. Symbolic Interactionism. Perspective and
Method. Englewood Cliffs, New Jersey: Prentice-Hall, Inc.

Bunge, Mario. 1999. The Dictionary of Philosophy. Amherst, New York:
Prometheus Books.

Edwards, Cliff. 1999. "Many Products Have Gone the Way of the Edsel."
Johnson City Press, 23 May 1999, 28, 30.

Glaser, Barney G. 1978. Theoretical Sensitivity. Advances in the
Methodology of Grounded Theory. Mill Valley, California: Sociology
Press.

Glaser, Barney G. 1992. Emergence vs. Forcing. Basics of Grounded
Theory. Mill Valley, California: Sociology Press.

Glaser Barney G., and Anselm Strauss 1967. The Discovery of Grounded
Theory. Strategies for Qualitative Research.

Hilpinen, Risto. 1995. "Charles Sanders Peirce." The Cambridge
Dictionary of Philosophy. Robert Audi, editor. Cambridge, England:
Cambridge University Press, 565-568.

Lukas, Paul. 1998. "The Ghastliest Product Launches." Fortune, 16
March 1998, 44.

Mansfield, Edwin, J. Rapaport, J. Schnee, S. Wagner and M. Hamburger.
1971. Research and Innovation in Modern Corporations. New York:
Norton.

Mautner, Thomas. 1996. A dictionary of philosophy. Oxford: Blackwell.

McMath, Robert. 1998. What Were They Thinking? Marketing Lessons I've
Learned from Over 80,000 New Product Innovations and Idiocies. New
York: Times Business.

Nonaka, Ikujiro and Hirotaka Takeuchi. 1995. The Knowledge-Creating
Company: How Japanese Companies Create the Dynamics of Innovation.
Oxford: Oxford University Press.

Peirce, Charles S. 1877. "The Fixation of Belief." Popular Science
Monthly 12 (November 1877), 1-15.

Strauss, Anselm, and Juliet Corbin. 1990. Basics of Qualitative
Research. Grounded Theory Procedures and Techniques. Newbury Park,
California: Sage Publications.

Wood, John. 2000. "The culture of academic rigor: does design
research really need it?" The Design Journal, Vol. 3, Issue 1, 44-57.